A colleague was deleting million plus rows from a table , when others tried connecting to the Oracle database their sessions seemed to hang. Just trying to find a cause for this as I believe this shouldn't have happened. The deletion didnt use the nolog option and the archived redo logs were full , but go no errors in the alert file. The Undo tablespace reached its maximum and was unable to grow any further. Would any of these stop people connecting ? The cpu utilisation on the server was very low. The solution ended up killing oracle and restarting the server, but just trying to find a root cause. in the Listener.log, everything looked normal and was still accepting connections,and another user was doing a full table scan select on the table that was having rows deleted. But still no reason for users not being able to connect, even as sysdba locally
the archive log directory wasnt full, but very busy as you could imagine, the undo tablespace was full, is it more likely to be the undo or the archived log that caused the problem, if its the undo is it worth using undo limits in the resource plan?